English

A log-Sobolev inequality for the multislice, with applications

Probability 2018-09-12 v1 Discrete Mathematics Combinatorics

Abstract

Let κN+\kappa \in \mathbb{N}_+^\ell satisfy κ1++κ=n\kappa_1 + \dots + \kappa_\ell = n and let Uκ\mathcal{U}_\kappa denote the "multislice" of all strings uu in []n[\ell]^n having exactly κi\kappa_i coordinates equal to ii, for all i[]i \in [\ell]. Consider the Markov chain on Uκ\mathcal{U}_\kappa, where a step is a random transposition of two coordinates of uu. We show that the log-Sobolev constant ρκ\rho_\kappa for the chain satisfies (ρκ)1ni=112log2(4n/κi),(\rho_\kappa)^{-1} \leq n \sum_{i=1}^{\ell} \tfrac{1}{2} \log_2(4n/\kappa_i), which is sharp up to constants whenever \ell is constant. From this, we derive some consequences for small-set expansion and isoperimetry in the multislice, including a KKL Theorem, a Kruskal--Katona Theorem for the multislice, a Friedgut Junta Theorem, and a Nisan--Szegedy Theorem.

Cite

@article{arxiv.1809.03546,
  title  = {A log-Sobolev inequality for the multislice, with applications},
  author = {Yuval Filmus and Ryan O'Donnell and Xinyu Wu},
  journal= {arXiv preprint arXiv:1809.03546},
  year   = {2018}
}
R2 v1 2026-06-23T04:01:26.114Z